The developing T-cell compartment of the neonatal lung orchestrates an atypical response to respiratory syncytial virus

1 The Institute of Virology and Immunology, Mittelhäusern and Bern, Switzerland 2 Department of Infectious Diseases and Pathobiology, Vetsuisse Faculty, University of Bern, Bern, Switzerland 3 Graduate School for Cellular and Biomedical Sciences, University of Bern, Bern, Switzerland 4 Institute of Animal Pathology, Department of Infectious Diseases and Pathobiology, Vetsuisse Faculty, University of Bern, Bern, Switzerland 5 COMPATH, Vetsuisse Faculty & Faculty of Medicine, University of Bern, Bern, Switzerland


32
The human respiratory syncytial virus (RSV) is a seasonal virus, known as the major 33 cause of lower respiratory tract infection during early childhood (1,2). Worldwide, RSV 34 leads to around 160'000 deaths each year (3). Although there are risk factors for RSV 35 severity such as prematurity and congenital heart disease, the majority of hospitalized 36 infants are previously healthy (4)(5)(6)(7), suggesting that the disease is partially linked to 37 inherent properties of early life immunity. The precise understanding of the immune-38 driven susceptibility of neonates is essential, considering there is no licensed vaccine 39 and available prophylactic treatments are mostly based on palivizumab, a neutralizing 40 antibody used in high-risk group infants (8). 41 The first exposure to pathogens occurs during the early postnatal period and is critical 42 for lung colonization by immune cells. There is increasing evidence that the overall 43 neonatal T-cell compartment is distinctive during this temporal window (9). Upon 44 stimulation neonatal T-cells tend to differentiate into regulatory T-cells (Tregs), to 45 facilitate self-tolerance to developing organs (10). Also, the developing lung is 46 characterized by a type 2 immune phenotype, and the pulmonary T-cell responses 47 present a T helper type 2 (Th2) bias (11,12). Remarkably, neonatal gd T-cells, a T-48 cell subset enriched at mucosal barrier sites, have an impaired capacity to respond to 49 stimulation (13)(14)(15)(16). 50 Due to the aforementioned reasons, young infants do not mount a classical pulmonary 51 immune response to pathogens such as RSV, which may contribute to high morbidity 52 and mortality (17,18). A likely explanation could be a combined contribution of viral 53 load (19) and an inappropriate and/or dysfunctional immune response, through an 54 over-exuberant inflammatory response or a biased T-cell response (9,20,21). 55 Neonatal T-cells may have both protective and harmful effects in responses to RSV 56 6 The primer and probe sequences used to detect RSV have been previously described 107 (32). 108 109

Flow cytometry 110
The different ovine immune cell subtypes were identified by flow cytometry (FCM) 111 using an eight-step, seven-color staining protocols. All antibodies used for the 112 procedure, as well as their clones, host, working dilution, and references, are listed in 113 Table E1. Combination stainings analyzed plasmacytoid dendritic cell (pDC), gd T-114 cells, Tregs, CD4 and CD8 T-cell subsets. For the acquisitions, 10 6 events were 115 accumulated for each sample. The experimental schedule is summarized in Table E2. 116 For the counts of cell subtypes, we calculated the ratio (number of events in the gated 117 cell subtype) to (number of all events, excepting cell aggregates and debris

Respiratory disease following neonatal RSV infection 133
Selected time-points were as follow: 3 and 6 days post-infection (p.i.), to evaluate the 134 innate and early adaptive responses; 14 days p.i. to assess the adaptive mechanisms 135 and virus clearance; 42 days p.i. to investigate the impact of RSV infection on the 136 immune cell colonization of the developing lung. As controls, non-infected newborns 137 and adults were used ( Figure 1A). 138 We quantified RSV in BALs (cellular and fluid fractions) and in the lung tissue. 139 Although RSV clearance was slower in the cellular fraction of BALs, it became 140 undetectable in any sample at day 42 p.i., demonstrating an efficient clearance ( Figure  141 1B). Immunohistochemistry of lung tissue sections revealed the presence of RSV-142 antigens in type I pneumocytes and macrophages ( Figure 1C). 143 Clinically, RSV infected neonates showed signs of rhinitis with clear nasal discharge, 144 occasional coughing and wheezing. Macroscopical lesions were visible at days 3-6 145 p.i, including failure of pulmonary collapse and focal, dark-red areas of subpleural 146 pulmonary consolidation and atelectasis. Histological lesions, present at days 3-6 p.i 147 consisted of multifocal-coalescing areas of lymphocytic and histiocytic infiltrates within 148 interalveolar septae (interstitial pneumonia) and accumulation of inflammatory cells 149 within alveolar spaces and bronchioles (bronchiolitis) ( Figure 1D). Longitudinal 150 measurements of white blood counts and neutrophil counts in the circulation showed 151 no difference between mock and infected neonates ( Figure E1). 152 153

Neonatal RSV infection activates the initiators of the adaptive immune response 154
When quantified in the BALs, pDC numbers of mock neonates were similar to those 155 of healthy adults. Upon RSV infection, we observed a trend to pDC recruitment as 156 early as 3 days p.i., and this was enhanced at day 6 p.i. (p<0.05). However, this pDC 157 recruitment was transitional and tightly regulated, since day 14 p.i corresponded to a 158 return to steady-state levels (Figure 2A,B). One out of six infected neonates failed to 159 display any pDC recruitment; this animal was the only one for which RSV was 160 undetectable in the fluid fraction of BAL ( Figure 1B). To verify whether the extent of 161 pDC recruitment was linked to the magnitude of RSV shedding, pDC counts were 162 plotted as a function of RSV copies in the fluid fraction of BALs. A significant 163 association was found ( Figure 2C). We next evaluated if the recruited pDCs displayed 164 a mature and activated phenotype. A significant CD86 upregulation was found at days High frequencies of IL-4-expressing CD8 + T-cells (Tc2) were also found in mock 207 neonates. Again, a progressive decline of Tc2 cells occurred up to day 42 p.i., 208 comparable to negligible levels measured in healthy adults. RSV disease induced a 209 greater overall Tc2-derived cytokine environment in BALs. This exacerbation was 210 tightly regulated, reduced to a level similar to healthy adults at day 14 p.i. Again, no 211 consistent influence of RSV infection was seen on IFN-g induction ( Figure 3E,F). 212 Altogether, these results highlight the role of Tc2 cells as an important source of IL-4 213 production during neonatal RSV infection. 214 We then tested the activation and differentiation of naïve T lymphocytes into memory 215 subsets. The most striking observation was achieved at day 42 p.i.; the percent 216 contributions of different subsets seen in RSV-infected neonates was highly 217 superimposable to that found in healthy adults. This was not the case for mock 218 neonates who had lower frequencies of activated memory T-cells, particularly for the 219 CD8 + fraction ( Figure 3G). Consequently, RSV infection in early life modulates the 220 activation and the maturation of the T-cell pool. 221 Next, we monitored the presence of RSV neutralizing antibodies (NAb) (Figures 3H,222 E2). In the serum of healthy adults and mock neonates, no RSV neutralizing antibodies 223 were detected. In contrast, RSV-infected neonates have naturally acquired NAb at day 224 13-14 p.i., except one animal. Notably, two siblings had RSV NAb preceding infection; 225 this was passively transmitted by their mother whose serum also had a strong RSV-226 neutralization ( Figure E3). We extended the assay to later time points to test the 227 persistence of RSV neutralization. No loss of humoral memory was measured up to 228 42 days p.i ( Figure 3I). Altogether, these results indicate that despite an inappropriate 229 cell-mediated immune response, neonates can clear RSV by mounting an efficient 230 humoral response. 231

Neonatal RSV infection rapidly induces Treg suppressive functions to dampen 233 the Th2 and Tc2 responses 234
The shortened kinetics observed for CD4 + T-cells was elaborated by determining 235 whether this was due to an early Treg rather than a classical CD4 + T-cell response 236 ( Figure 4A). RSV infection elicited a rapid Treg recruitment in BALs that mirrored the 237 one previously observed for CD4 + T-cells. This Treg dynamic is specific for the airway 238 mucosa, since Treg frequencies remained unchanged in the lung tissue ( Figure 4B). 239 We then noticed a trend for Treg subset overrepresentation following RSV infection 240 ( Figure 4C). The proportion of Tregs declined progressively over time, tending towards 241 similar levels to those detected in healthy adults. RSV infection promoted 242 enhancement of TGF-β production by Tregs, however not significant ( Figure 4D). 243 When Treg counts were calculated as a function of non-Treg CD4 + counts, an 244 association was found, suggesting that CD4 + T-cell pool tends to differentiate into 245 Tregs. When Treg counts were calculated as a function of CD8 + counts, an association 246 was still found, confirming that CD8 + T-cells and Tregs colonized the alveolar space 247 in a coordinated manner. By plotting CD8 + count as a function of non-Tregs CD4 + 248 counts, it appeared that T-cell response was mainly driven by the CD8 + fraction (high 249 numbers at days 6 and 14 p.i.) ( Figure 4E). In peribronchial LNs following RSV 250 infection, CD8 + counts remained stable over the course of the disease, showing that 251 their accumulation is airway mucosa-specific ( Figure 4F). Moreover, unlike the 252 situation described for the BAL compartment, RSV infection did not promote a further 253 increase of Tc2 cells ( Figure 4G). 254 255 Distinct cellular immune response in neonates compare to adults following RSV 256 infection 257 We followed the same approach to dissect the cellular immune response in RSV-258 infected adults. While the virus clearance was comparable between adults and 259 neonates ( Figure 5A), respiratory lesions were absent in adults ( Figure 5B). 260 The rapid and robust pDC recruitment observed in infected neonates did not occur in 261 infected adults. However, lung-resident pDCs of the latter were prompt to react, 262 displaying an even more mature and activated phenotype Importantly, this mature 263 pDC response was also tightly regulated ( Figure 5C,D). 264 The strong decrease in gd T-cell counts observed in infected neonates was absent in 265 adults. Moreover, adult gd T-cells produced high amounts of IL-17, suggesting that IL-266 17-mediated immunity is deficient in early life ( Figure 5E,F) (15). This IL-17 response 267 was accompanied by an activated phenotype, as indicated by the enhanced proportion 268 of CD25 + cells ( Figure 5G). From these results, we could ascertain that neonates lack 269 a mature and functional gd T-cell pool in the lower airways. 270 For the CD4 + compartment, no significant recruitment was found ( Figure 5H). The 271 percentage of IL-4-producing cells was very low, and RSV infection had no consistent 272 influence on IFN-g induction ( Figure 5I). Another clear difference was the contraction 273 of effector memory pool (CD25 + CD45RO + ) at day 14 p.i., indicative of a specific 274 immune response ( Figure 5J). 275 The recruitment of CD8 + T-cells was delayed compared to neonates, only detectable 276 at day 14 p.i. ( Figure 5K). Again, the percentage of IL-4-producing cells was low 277 ( Figure 5L), whereas the precipitous contraction of effector memory pool at day 14 p.i. 278 suggested a RSV-specific CD8 + T-cell response ( Figure 5M). From these results, it 279 13 became clear that the inaptitude of neonates to mount an effector T-cell response is 280 tentatively substituted by a fast and massive Th2/Tc2 cell recruitment. 281 Finally, we failed to detect any Treg recruitment. However, the rare tissue-resident 282 Tregs were able to produce TGF-β, suggesting that an immune regulation happened 283 ( Figure 5N,O). The pulmonary T-cell response was evaluated in neonates infected with RSV-ON1-288 H1 strain, which induced a milder disease ( Figure 6A). We did not observe a gd T-cell 289 depletion, and the CD8 + T-cell recruitment was delayed. Furthermore, the Treg 290 recruitment was lower compared to RSV A2 infection ( Figure 6B). By combining 291 results obtained with both strains, we found a significant association between Treg 292 and pDC counts in BALs ( Figure 6C). This indicates that, despite their skew, all distinct

298
To properly assess the immunopathogenesis of RSV, it is essential to select a model 299 that recapitulates the features of human disease. We selected the neonatal lamb, as 300 RSV is a natural pathogen for this species, lung development is similar, and the key 301 features mimic human RSV infection in infants (27, 28). One strength of our 302 experimental approach was the ease sampling and the full accessibility to the lower 303 airways throughout the disease when sampling in infants is mainly restricted to the 304 upper airways, the latter representing only an approximate surrogate of the distal lung 305 where RSV disease occurs. Moreover, we offer an unprecedented view of the 306 development of pulmonary neonatal immunity in healthy conditions. This process is 307 highly dynamic and tightly regulated, with a short and early stage where colonizing T-308 cell subsets synergize towards a narrow pro-tolerogenic window, namely a 309 "Treg/Th2/Tc2" environment. This certainly contributes to the maturation of the 310 immune system and may constitute an unsuitable state to resolve RSV infection. 311 The mechanisms underlying the severity of disease are still unclear. Herein, we 312 observed a comparable virus clearance between adults and neonates, reinforcing the 313 assumption of an inappropriate immune response. We observed a rapid and robust 314 recruitment of neonatal pDCs, whose extent correlated with the viral loads of infected 315 neonates. Infected adults did not require this recruitment, since their lung-resident 316 pDCs reacted promptly by maturing to a higher level than that measured in neonates. 317 This partially corroborates recent studies reporting low counts and maturation levels 318 of pDCs in mice and infants (33, 34). This reduced capacity to respond against RSV 319 might be due to a direct disturbance of signaling by RSV itself (35), or an intrinsic 320 deficiency of neonatal pDCs to produce elevated amounts of cytokines (36-38). This 321 neonatal defect would prevent conventional DCs from adequate presentation of viral 322 antigens, contributing to poor specific T-cell responses (39). 323

Controversial is whether gd T-cells participate in viral clearance or immunopathology. 324
Herein, the gd T-cell depletion conflicted with a previous report. Indeed, children with 325 severe RSV bronchiolitis had reduced frequencies of gd T-cells in peripheral blood, 326 which was explained by the authors as a likely redistribution towards the lungs; 327 unfortunately, they could not evaluate their hypothesis (40). In line with our findings, 328 mice depleted of gd T-cells before RSV infection display increased viral titers (13). An 329 alternative hypothesis that would reconcile our study with others, is that gd T-cell tissue 330 redistribution plays less of a role than its immature state in early life. Peripheral blood 331 gd T-cells of infants with severe disease failed to produce IFN-g when restimulated 332 (14), whereas gd T-cells from neonatal mice had an impaired ability to produce IL-17 333 (15). Finally, Tc17 and Th17 cells were recently associated with shorter hospital stays 334 and proposed to play a protective role (26). Altogether, our data demonstrate that the 335

protective role of gd T-cells in RSV infection is largely inefficient in early life. 336
The BALs of neonates displayed a high proportion of Tregs that declined rapidly to 337 levels comparable to healthy adults. As fetal CD4 + T-cells tend to differentiate 338 preferentially into Tregs after stimulation (10), this tolerogenic mechanism is likely to 339 promote self-tolerance to the developing lung. We found that Treg response to RSV 340 infection happened very early in neonates, accompanying CD4 + and CD8 + T-cell 341 expansion rather than shutting them down afterwards. This strong Treg recruitment 342 was absent in adults. We speculate that the neonate immune system faces a dilemma 343 between avoiding potential autoimmune disorders and failing to mount a specific T-344 cell response to harmful pathogens. Studies conducted in young infants with RSV 345 disease support this statement. A selective depletion of peripheral Tregs was shown, 346 probably due to massive recruitment to the lungs (41). A higher level of TGF-β 347 transcript was measured in neonatal DCs compared to adults (42). In contrast, some Herein, not only we show that this happens, but we also unify the link between IL-4 382 and RSV disease and endogenous IL-4 program in neonates. This finding constitutes 383 an important step towards deciphering why neonates and young infants are 384 particularly at risk.